AI Observer
A self-hosted, single-binary observability backend for local AI coding assistants that tracks token usage, cost, latency and error rates across tools such as Claude Code, Gemini CLI and OpenAI Codex CLI in one dashboard.
AI Observer ingests telemetry from local AI coding tools, either by watching each tool's session files directly with no OpenTelemetry configuration needed, by importing past session files with cost calculated after the fact, or by receiving standard OTLP data over HTTP in JSON or protobuf form. A real-time dashboard, updated over WebSocket, shows token usage, API latency, error rates and session activity, with a drag-and-drop builder for saved widgets.
Embedded pricing data covers Claude, Codex and OpenAI, Gemini and GitHub Copilot model aliases, and OpenCode's own plugin supplies its cost metrics natively. Storage and queries run on DuckDB for fast analytics over large datasets, and the whole application ships as one roughly 54MB binary with an embedded frontend, or as small multi-architecture Docker images, so no external services are required to run it locally.